Close-up eye portraits are everywhere on Instagram right now — a face filling the whole frame, the eyes sharp, loose hair strands crossing over, and the background falling away into shadow. The look depends almost entirely on how the prompt describes the light and the skin, which is why the three prompts on this page give three very different results from the same idea.
All three are complete and ready to copy. Each one sits under the result it produces, so you can see what you are getting before you copy it.
Prompt 1 — Eyes Above a Dupatta
This is the longest and most controlled of the three. It describes the framing, the makeup, the fabric and the lighting section by section, and it ends with a negative prompt that blocks the usual AI failures such as plastic skin and distorted eyes. Use it when you want a result close to the example and you do not want the tool improvising.
Copy Prompt ⤵️
✨ AI Prompt 1
Create an ultra-realistic close-up portrait photograph of a young South Asian woman, framed extremely tightly around her eyes, eyebrows, upper nose, forehead, and partially covered lower face.
COMPOSITION:
Extreme close-up portrait, face filling almost the entire frame. The camera is positioned directly in front of her face at approximately eye level. Both eyes are clearly visible and centered horizontally. Only the upper portion of the face is exposed while the lower half is naturally covered by a dark traditional patterned dupatta/scarf. The framing should feel like a candid smartphone portrait captured very close to the subject.
FACE & EXPRESSION:
Natural youthful South Asian facial features, realistic skin texture, soft forehead, naturally shaped thick dark eyebrows, large expressive dark-brown eyes looking directly into the camera. Eyes should be the main focal point. Subtle natural under-eye texture and slight redness around the inner corners of the eyes. No exaggerated expression, calm and slightly mysterious gaze.
EYES & MAKEUP:
Dark brown realistic irises with detailed natural reflections and visible catchlights. Thick black eyelashes with subtle mascara. Soft blue/grey eyeliner or eyeshadow tracing the upper eyelids, slightly winged toward the outer corners. Very subtle shimmer around the inner eye area. Natural lower eyelashes clearly visible. Makeup should look realistic and slightly imperfect rather than digitally polished.
HAIR:
Long dark brown/black hair, naturally straight with a slight soft texture. Hair falls from the upper-right side of the frame toward the forehead and temple. Several extremely fine loose strands of hair cross naturally over the forehead, eyes, and bridge of the nose. Preserve realistic individual hair strands.
SKIN:
Warm medium-light South Asian skin tone with realistic pores, fine skin texture, subtle natural imperfections, soft peach undertones, and gentle highlights on the forehead and nose. Absolutely no plastic skin, excessive smoothing, artificial beauty filter, or over-retouching.
CLOTHING:
The lower half of the face is covered by a traditional dark navy/black dupatta or scarf with an intricate antique-gold paisley and floral embroidered pattern. The fabric should appear rich, slightly textured and woven, with detailed golden motifs and borders. The dupatta is held naturally close to the face and occupies the lower portion of the image.
LIGHTING:
Warm indoor lighting, soft but slightly direct illumination on the face. A subtle warm flash-like highlight creates bright realistic catchlights in both eyes and gentle highlights on the forehead and nose. Soft shadows around the eyes and fabric folds. Natural warm golden-brown color temperature.
CAMERA:
Shot on a high-quality smartphone or professional portrait camera, approximately 85mm equivalent portrait perspective, very shallow depth of field, sharp focus specifically on both eyes and eyelashes, realistic optical falloff toward the surrounding areas. Natural photographic depth, no artificial CGI appearance.
COLOR GRADING:
Warm cinematic tones, slightly muted contrast, rich dark blacks, subtle golden highlights, natural skin tones, deep navy fabric, antique gold embroidery. Slight vintage/film-like warmth while maintaining realistic skin colors.
DETAIL:
Photorealistic, highly detailed eyes, individual eyelashes, individual hair strands, realistic fabric weave, embroidery details, natural pores and skin texture, realistic reflections in the eyes, subtle lens softness, authentic smartphone photography aesthetic.
STYLE:
Ultra-realistic candid portrait photography, intimate close-up, natural beauty photography, cinematic realism, authentic South Asian aesthetic, subtle warm film photography look.
ASPECT RATIO:
4:5 vertical portrait.
NEGATIVE PROMPT:
cartoon, anime, illustration, CGI, 3D render, plastic skin, airbrushed skin, excessive beauty filter, over-smoothed face, unrealistic eyes, extra eyes, distorted face, asymmetrical eyes, fake eyelashes, exaggerated makeup, oversaturated colors, excessive HDR, artificial lighting, blurry eyes, deformed facial features, duplicate facial features, unnatural hair, excessive bokeh, low resolution, watermark, text, Instagram interface, social media icons
Prompt 2 — Macro Single Eye in Golden Light
This one is built around a strict identity lock: it tells the tool to keep your exact eye shape, iris colour, lashes and skin texture, and only change the light and the framing. It is the shortest of the three and the best choice if you are uploading your own photo and want it to still look like you.
Copy Prompt ⤵️
✨ AI Prompt 2
Use the uploaded reference image. STRICT EYE IDENTITY LOCK: Preserve 100% the exact same eye – identical eye shape, iris color, pupil, eyelids, eyelashes, eyebrow, skin texture, pores, and facial proportions. Do not beautify or alter any eye details.
Ultra-realistic cinematic macro close-up of a single deep brown eye filling the frame, warm golden-hour sunlight, natural glossy catchlights, soft nude eye makeup, thin black winged eyeliner, thick natural eyebrow, ultra-detailed skin with realistic pores, a few loose hair strands crossing the face, hand softly covering the lower face, shallow depth of field, luxury beauty editorial photography, Canon EOS R5, 85mm lens, f/1.4, HDR, photorealistic, 8K, ultra-sharp, masterpiece, true-to-life colors.
Negative Prompt: changed eye, different iris, different eyebrow, altered eyelashes, beautified eye, extra eye, blurry, soft focus, plastic skin, cartoon, anime, CGI, low quality, low resolution, distorted face, AI artifacts, watermark, text, logo.
Prompt 3 — Night Flash Candid Portrait
The third prompt recreates the late-night phone-flash photo: harsh direct flash on the face, everything behind it in darkness, messy hair across the cheeks, and no attempt at a polished finish. It asks for 9:16, so it fits Stories and Reels covers directly.
Copy Prompt ⤵️
✨ AI Prompt 3
Use the uploaded reference image as the ONLY identity reference. STRICT IDENTITY LOCK – preserve the person’s exact face, facial structure, eyes, eyebrows, nose, lips, skin tone, skin texture, proportions and natural imperfections. Do not beautify, reshape, smooth, or alter the face.
Create an ultra-realistic intimate close-up portrait of the same young woman, photographed from a slightly tilted/sideways angle while she is resting her face against a soft light-colored pillow or fabric. The composition should be extremely close, showing mainly her forehead, both eyes, eyebrows, nose, cheeks and lips, with part of the face extending naturally outside the frame.
She has deep brown expressive eyes, naturally thick eyebrows, subtle natural eyelashes, realistic skin pores and fine facial texture. Her expression is soft, quiet, slightly emotional and dreamy, looking directly toward the camera. Keep her natural facial asymmetry and realistic features.
Dark brown/black hair, naturally messy and slightly tousled, falling loosely around the forehead and cheeks. Several very fine individual hair strands naturally cross over the forehead, eyes, nose and cheek, creating an authentic candid look. Hair should have realistic individual strands and subtle shine.
Use direct smartphone camera flash photography, creating a warm, slightly harsh flash illumination on the face while the surrounding background falls into deep darkness. Natural warm brown skin tones, realistic highlights on the eyes and nose, subtle flash reflection/catchlights in both eyes.
The image should feel like an unposed late-night Instagram/Reels aesthetic portrait, intimate, raw and candid rather than professionally posed. Slight smartphone-camera imperfections, realistic exposure, subtle grain, natural shadows, high skin detail, no artificial beauty filter.
Composition: extreme close-up, face filling almost the entire frame, slightly rotated/diagonal framing, eyes positioned around the center, shallow depth of field, background almost completely dark and indistinct.
Lighting: single direct smartphone flash, warm natural skin illumination, soft falloff, deep shadows, realistic specular highlights, authentic low-light photography.
Camera: smartphone rear camera, close focusing distance, realistic smartphone optics, subtle computational photography, natural HDR, realistic dynamic range, fine photographic grain.
Style: ultra-photorealistic, candid Instagram aesthetic, intimate bedroom/night photography, raw natural beauty, realistic skin pores, realistic hair strands, cinematic but NOT overly polished, high detail, 4K, photorealistic.
Aspect ratio: 9:16 vertical.
IMPORTANT: No face modification, no makeup enhancement, no skin smoothing, no facial reconstruction, no exaggerated eyes, no artificial symmetry, no plastic skin, no beauty filter, no extra jewelry, no text, no watermark, no Instagram interface.
Which Prompt Should You Use?
Prompt 1 — most control, most detail. Best when you want the dupatta framing and studio-style warmth. It is long, so copy all of it.
Prompt 2 — strongest identity lock. Best when the eye in the result must still be recognisably yours.
Prompt 3 — most natural looking. Best for the raw, unposed phone-flash aesthetic that suits Reels and Stories.
All three ask for realistic skin with visible pores and no beauty filter. That is deliberate: the reason these portraits look real is the texture, and removing those lines is the fastest way to get a plastic result.
Which Photo to Upload
Prompts 2 and 3 work from your uploaded photo, so the upload decides how close the result stays to you. Prompt 1 can work with or without a reference.
Use a sharp photo where the eyes are in focus. Everything in these prompts centres on the eyes.
Good light on the face. A dark source photo gives the tool nothing to lock on to.
No existing beauty filter. The AI copies the smoothing and the skin ends up looking artificial.
For Prompt 3, a photo taken fairly close to the face matches the framing better than a distant shot.
Step by Step
Open an AI image tool that accepts a photo plus text — ChatGPT with image generation, Google Gemini or Grok.
Upload your photo first, then paste the prompt underneath it.
Copy the prompt in full, including the negative prompt at the end where there is one. The negative lines are what stop the plastic-skin look.
Generate, then download the largest size available.
Run it a few times. Close-up faces vary more between attempts than wide shots do.
Fixing the Problems That Come Up Most
The eyes changed shape or colour. Use Prompt 2, which locks identity, or add preserve the exact eye shape and iris colour from the uploaded photo to whichever prompt you are using.
The skin looks plastic. Add visible skin pores, natural texture, no smoothing. This is the single most common failure in close-up portraits.
The hair strands disappeared. Add keep individual loose hair strands across the face. Tools tend to tidy hair away unless told not to.
The lighting looks flat. For Prompt 3 add single direct flash, deep dark background; for Prompt 2 add warm golden-hour backlight.
You got a square image. Repeat the aspect ratio in plain words: 4:5 for Prompt 1, 9:16 for Prompt 3.
The face came out symmetrical and doll-like. Add keep natural facial asymmetry and small imperfections.
Where These Work Best
Instagram posts — Prompt 1 at 4:5 fills the feed properly.
Stories and Reels covers — Prompt 3 at 9:16 needs no cropping.
Profile pictures — Prompt 2 crops well to a circle because the eye sits near the centre.
Phone wallpapers — Prompt 3, with the dark background keeping icons readable.
It describes each part of the image separately — composition, eyes, hair, skin, fabric, lighting, camera, colour. That length is what makes it repeat reliably. Shorter prompts leave more to the tool, which means more variation between runs.
What is a negative prompt and do I need it?
A negative prompt lists what you do not want: cartoon, plastic skin, extra eyes, watermarks. Not every tool supports a separate negative field, but pasting those lines along with the prompt still helps in most cases.
Can I use a photo of someone else?
Only with their permission. The photo you upload is sent to the AI service you use, so check that service’s terms before uploading anyone else’s picture.
Why do my eyes look slightly different each time?
AI tools do not repeat an exact result. For the closest match to your own eye, use Prompt 2 and generate three or four times before choosing.
Conclusion
Three prompts, three different looks: a controlled studio close-up, a macro eye with a hard identity lock, and a raw night-flash candid. Copy whichever matches what you want, upload a sharp well-lit photo, and keep the skin-texture and negative-prompt lines intact — those are doing more work than any other part of these prompts.
